How to pass 5D tensor to nn.Linear() function

1

Hello all, I have a 5D tensor such as BxCxHxWxD. I want to use nn.Linear function to classify from C to C//2 channel. How should I do it?

This is my code

class 5D_Linear(nn.Module):
    def __init__(self, channel):
        super(5D_Linear, self).__init__()
        self.avg_pool = nn.AdaptiveAvgPool3d(1)
        self.linear_5d = nn.Sequential(nn.Linear(channel, channel// 2),
                                                 nn.ReLU(inplace=True),
                                                 nn.Linear(channel // 2, channel))


    def forward(self, x):
        x = self.avg_pool(x)
        B, C, D, H, W = x.size()       
        x_4d = x.view(B, C, D, -1)
        x_4d = self.linear_5d(x_4d)
        x_5d = x_4d.view(B, C, D, H, W)
        print(x_5d.size())

        return x_5d
2

linear_5d seems to be mapping channel//2 back to channel , is that your intention? what doesn’t seem to be working in this code?

3

sorry. this is my mistake. the 5d must convert to 1d to perform classification. it looks compute prob of each channel